Develop energy-based neural solvers that embed memory effects to accelerate geophysical fluid dynamics simulations. Combine PDE-based modeling, computer graphics methods and deep learning to enable high-resolution, computationally efficient simulations for climate, hazard and landscape evolution.